The Financial Early Warning Research Of Based On The Least Squares And Data Envelopment Analysis

In the context of the global economic crisis, the financial market in China is influenced to some extent, each enterprise must be ensured to accurately predict the risks they may face in the future, it is the only way by which enterprises can gain advantages in the global economy. So, the managers of enterprises have attached great importance to the financial early-warning, which can make enterprises avoid risks in the operation and reap the benefits. Managers of enterprises and scholars believe that the financial early-warning has very important practical significance.In academia, research on the financial early-warning originated in 1932, the research is one discriminant model, then some scholars began to study the theory of financial early-warning, after constant supplement and perfection, the financial early-warning models are more accurate to predict the crisis of the enterprises. Until now: scholars have put forward the following financial early-warning models: one discriminant model, multiple discriminant model, multivariate logistic model,multivariate probability model, artificial neural network, support vector machine model and collaborative forecasting model, etc. Although these models have higher prediction accuracy, there are also some shortcomings.In order to make up the shortages of the previous models, in this paper, the panel data is used to construct the financial early-warning model, the random effects model of financial early-warning is established by using the least squares, the non-ratio index is cited in the model to enrich the range of index selection, and every index has higher influence coefficient. In order to make the enterprises in crisis get rid of crisis,this paper adopts the data envelopment analysis method, the method is good for the enterprises in crisis, through which the problems can be found, and the impact factors of enterprise in the crisis can be adjusted to reach the target value required in the data envelopment analysis software, so the enterprises can make adjustments accordingly to get out of trouble earlier. This paper selects 119 companies as the research samples;these 119 companies cover all types of enterprise, and have the characteristic of universality. The model also has a certain degree of prediction accuracy, and the prediction accuracy reached 95.24% and 95.24% respectively. Model also has the ability to adjust the wrong, more favorable to the enterprise out of the woods. Model also has the ability to adjust the wrong, and it has beneficial effect on enterprise out of the woods. But the model’s calculation process is complicated; it is not easy to operation.